Lab on a Chip

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

PAPER View Article OnlineView Journal | View Issue

Lab ChipThis journal is © The Royal Society of Chemistry 2014

a Biomaterials Innovation Research Center, Division of Biomedical Engineering,

Department of Medicine, Brigham and Women's Hospital, Harvard Medical

School, Boston, MA 02115, USA. E-mail: alik@rics.bwh.harvard.edubHarvard-MIT Division of Health Sciences and Technology, Massachusetts

Institute of Technology, Cambridge, MA 02139, USAcWyss Institute for Biologically Inspired Engineering at Harvard University,

Boston, MA 02115, USAdDepartment of Biomedical Engineering, College of Health Science,

Korea University, Seoul 136-100, Koreae KU-KIST Graduate School of Converging Science and Technology,

Korea University, Seoul 136-701, Republic of KoreafWPI-Advanced Institute for Materials Research (WPI-AIMR), Tohoku University,

2-1-1 Katahira, Aoba-ku Sendai 980-8577, JapangCollege of Animal Bioscience and Technology, Department of Bioindustrial

Technologies, Konkuk University, Hwayang-dong, Kwangjin-gu, Seoul 143-701,

Republic of KoreahDepartment of Maxillofacial Biomedical Engineering and Institute of Oral

Biology, School of Dentistry, Kyung Hee University, Seoul 130-701,

Republic of Koreai Department of Physics, King Abdulaziz University, Jeddah 21569, Saudi Arabia

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4lc00510d‡ Current address: Division of Mechanical Design Engineering, ChonbukNational University, Jeonju, 561-756 Korea

Cite this: Lab Chip, 2014, 14, 4213

Received 30th April 2014,Accepted 31st July 2014

DOI: 10.1039/c4lc00510d

www.rsc.org/loc

Siphon-driven microfluidic passive pump with ayarn flow resistance controller†

Gi Seok Jeong,ab Jonghyun Oh,‡a Sang Bok Kim,abc Mehmet Remzi Dokmeci,ab

Hojae Bae,g Sang-Hoon Leede and Ali Khademhosseini*abcfhi

Precise control of media delivery to cells in microfluidic systems in a simple and efficient manner is a

challenge for a number of cell-based applications. Conventional syringe pumps can deliver culture media

into microfluidic devices at precisely controlled flow rates, but they are bulky and require a power source.

On the other hand, passive microflow-generating systems cannot maintain continuous, controllable and

long-term delivery of media. We have developed an on-chip microflow control technology that combines

flow rate control and passive, long-term delivery of media to microwell tissue culture chambers. Here, a

passive flow is initiated using the siphon effect and a yarn flow resistor is used to regulate the flow

rate in the microchannel. Using the yarn flow resistor, the medium flow rate into the microfluidic cell

culture system is made adjustable to a few hundred microliters per hour. To evaluate the effects of

controlled flow on microfluidic cell culture properties (feasibility test), we measured the cell alignment

and cytoskeletal arrangement of endothelial cells cultured in a microwell array inside the microfluidic

channel.

Introduction

Continuous and controlled delivery of culture media to cellsin microfluidic chips is essential in biomedical applicationsassociated with drug delivery,1 chemical gradient genera-tion,2,3 interstitial flow-driven cell migration,4 or oxygen andnutrient delivery to organs-on-a-chip.5 The ability to control

the flow rate in a cell culture chip within a specific range iscrucial because the flow rate can affect the metabolicexchange and shear stress applied to the cells during culture.The interstitial flow (10−1–100 μm s−1) can increase focal adhe-sion kinase activation, which is a key factor for migration,sprouting, and alignment among cancer cells.6 Interstitialflow can also generate protein gradients7 and deliver diversecytokines.8 Furthermore, the flow shear stresses mimickingin vivo conditions (10−1–101 dyne cm−2) can stimulate stemcell differentiation9–11 and cell alignment.12–15 Shear stressalso plays an important role in the cytoskeletal reorganizationof endothelial cells (ECs),12,13,15 nitric oxide syntheses,16 andapoptosis inhibition.17 Therefore, generation of microflowwith a flow rate similar to that of physiological-level flow(10−1–101 dyne cm−2) in microfluidic channels without anyuse of peripheral devices is critical for mimicking the in vivomicroenvironment.

Syringe pumps are commonly used to generate and regu-late microflows.18,19 However, continuous delivery of mediainto microfluidic devices within the cell culture chambersremains challenging since it requires large pumps and apower source. Passive microflow-generating systems, includ-ing osmotic pressure-driven micropumps,2,3 mediumreservoir-based flow control systems,4 and droplet-basedsurface tension-driven micropumps,20,21 are also frequentlyused. However the existing passive systems are limited toshort-term uses only, and flow rates are typically not finelyadjustable in these systems.

, 2014, 14, 4213–4219 | 4213

Lab on a ChipPaper

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

View Article Online

In this paper, we have developed a passive micropumpthat generates a controllable flow rate similar to the intersti-tial flow. We employed the siphon effect to initiate the flowand a porous yarn to control the flow rate. The height differ-ence between the upper (inlet) and the lower (outlet) fluidreservoirs created a pressure difference (a driving force), andthe force applied by the weight of the water column inducedcontinuous flow. The passive pump system is composed oftwo reservoirs (inlet and outlet) and a microchannel betweenthe two reservoirs. The flow rate of the micropump was regu-lated by integrating a piece of yarn into the flow chamber.This yarn acts as a flow resistor to regulate the flow rate. Theflow rate was maintained within the range of a few hundredmicroliters per hour. For the feasibility test, we measured thecell alignment and cytoskeletal arrangement of endothelialcells cultured in a microwell array inside a microfluidic chan-nel in which the controlled flow rate was similar to that ofphysiological flow.

Materials and methodsPrinciple of the yarn flow resistor (YFR)-based micropump

A yarn flow resistor (YFR) is a flow resistor that utilizes a yarncapillary resistance for siphoning off media in the range of afew hundred microliters per hour. The flow rate in the micro-channels without the YFR (channel thickness: 100–300 μm) ismaintained in the range of 10–100 mL h−1. To maintain theflow rate at the range of a few microliters per day, YFR pro-vides a suitable flow resistance of a siphon-driven passivepump onto a microchannel. The siphon effect is generatedby the height difference between the upper (inlet) and lower(outlet) fluid reservoirs, and the fluid moves under theunequal pressures toward the equilibrium configuration22

(ESI† Fig. S1d). The siphon effect in the presence of the YFRcan be described mathematically using the expandedBernoulli's equation,

12

121

21

12

22

2V gZ P V gZ P gH L , (1)

where V, Z, and P indicate the velocity, position, and pres-sure, respectively. HL is the total loss of water height due tothe channel resistance, and g is the gravitational accelerationconstant. P1 and P2 were assumed to be equal to the atmo-spheric pressure, and the upstream velocity in the reservoirwas assumed to be relatively low compared to the velocity atthe outlet. Thus, P1, P2, and V1 can be neglected, and V2 ineqn (1) can be expressed as

V2 = [2g(ΔZ − HL)]1/2 (2)

Therefore, the velocity at the outlet depended on theheight difference (ΔZ), the resistance due to the YFR, and thechannel resistance (HL). The flow rate through a channel

4214 | Lab Chip, 2014, 14, 4213–4219

equipped with the YFR was determined according to the fol-lowing equation:

Q V D V= YFR outlet area =outlet YFR

2

24(3)

Thus, the flow rate through amicrofluidic chip with the YFRwas determined by calculating the pressure drop (HL). The experi-mentalmeasurements were used to calculateHL as shownbelow,

HL ∝ CYFR × length of YFR + base pressure drop, (4)

where CYFR is the pressure drop constant for the YFR andwas calculated to be 2.817 × 10−13 cm−1. Although the pres-sure increased with the YFR length, the base head drop con-tributed the largest amount to the YFR effects. The pressuredrop in the channel could be reduced by a few hundredmicroliters per hour using a short YFR by increasing theporous resistance of the YFR.

Fabrication of a microfluidic device and a siphon pump

The passive pump was fabricated in a simple, cost-effectivemanner within a 100 mm × 100 mm square dish. The inletreservoir and the microchannel are connected by a polyethyl-ene tube (OD = 1.27 mm, ID = 0.86 mm, Instech Laboratories,Inc., PA, USA), and the outlet reservoir and the microchannelare connected with the cotton YFR. The microfluidic channelwas fabricated using conventional soft lithography. Thepoly(dimethylsiloxane) (PDMS; Sylgard 184, Dow Chemical,MI, USA) channel was replicated from the master moldpatterned with SU-8 on the silicon wafer (MicroChem, MA, USA).After an oxygen plasma treatment (Femto Science, Korea), thereplicated PDMS was bonded to the cover glass.

YFR was fabricated using a cotton yarn (twisted cottonbutcher's twine (1 mm × 3 mm), Koch, USA) with a length ofapproximately 10 cm. YFR was inserted into the tube for eas-ier handling and connection to the microfluidic channel(ESI† Fig. S1a). To connect the reservoir to the microfluidicchannel, the tube containing YFR was bent slightly using analcohol lamp (ESI† Fig. S1b). To initiate siphon-driven flow, asyringe was connected to the outlet of the channel, and aflow was generated from the inlet reservoir to the outlet bydrawing up the air in the microchannel with a syringe at theoutlet (Fig. 1a and ESI† Fig. S1c). After removing the syringe,we connected the tube containing YFR to the outlet of themicrofluidic channel (Fig. 1a, b, and d and ESI† Fig. S1d).The flow initiated from the siphon effect was maintained bythe height (and therefore pressure) difference (ΔZ) betweenthe inlet and the outlet medium reservoirs (Fig. 1b) and thefluid flowing through the tubing (R1), the microchannel (R2),and the YFR (R3). Here, the YFR kept the fluid moving at aconsistent rate. The fluid flowed continuously through themicrochannel until the pressure difference between the inletand the outlet reservoirs equilibrated (ΔZ = 0). A conical tube(50 mL) was used to connect the inlet reservoir (Fig. 1d) and

This journal is © The Royal Society of Chemistry 2014

Fig. 1 YFR-regulated microflow control system. (a) Schematic diagram showing the siphon effects in the yarn capillary resistance-driven micro-pump system. (b) Schematic diagram of the water head difference-driven siphon effect controlled by the YFR. (c) Photographic image of a yarncapillary regulator. (d) The microfluidic device prepared with a YFR. The system consists of two fluid reservoirs: the microfluidic channels and theYFR (black arrowheads) as well as the drain reservoirs.

Lab on a Chip Paper

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

View Article Online

the drain tank, a 60mmdiameter Petri dish (Fig. 1d). Flow char-acteristics were evaluated by measuring the flow rate in a cellincubator to prevent the fluid evaporation from the cotton yarn.

Human microvascular endothelial cell (hMVEC) culture

hMVECs were purchased from Lonza (Basel, Switzerland) andwere cultured in endothelial cell medium (ECM; SciencellResearch Laboratories Inc., CA, USA) consisting of 500 mL ofbasal medium, 5% fetal bovine serum (FBS), 20 ng mL−1

VEGF, and 1% penicillin/streptomycin. The hMVECs wereexpanded for no more than seven passages. One hundredmicroliters of a hMVEC suspension (2 × 105 cells mL−1) wasseeded into the microfluidic chip, and the device was placedin a 37 °C incubator overnight to facilitate the attachment ofthe cells to the cover glass. After cell attachment, the deviceswere maintained at 37 °C in an incubator (5% CO2). The YFRwas connected to the microfluidic device, and the flow wasmaintained at approximately 3.5 mL per day to induce ashear stress on the hMVECs. The flow rate in the chip wasmaintained by replenishing the cell culture medium in thereservoir every other day. Cell responses were monitored dailyusing a phase contrast microscope (AMG, WA, USA).

Immunocytochemistry

Cells cultured over 48 hours under flow were fixed for 20 minwith 4% formaldehyde at 4 °C. For immunostaining, the cellswere permeabilized using 0.1% Triton X-100 in 0.1% PBS for20 min at room temperature and were blocked with 3% BSAin PBS for 30 min followed by incubation in the presence of a

This journal is © The Royal Society of Chemistry 2014

primary antibody overnight (4 °C). A primary antibody(Abcam, UK) against α-smooth muscle actin (1 : 1000) wasused to observe the cytoskeleton remodelling and cellalignments. After incubating overnight, the cells were washedwith PBST (0.05% Tween in PBS) for 5 min. Secondaryantibodies (1 : 1000 dilutions, Invitrogen, CA) were applied for1.5 hours at room temperature. Cells were washed again withPBS, and fluorescence images were acquired using a fluores-cence microscope (EVOS, AMG, USA) after counterstainingwith 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI,Invitrogen, CA). Cell alignment was measured using theImageJ software (http://rsbweb.nih.gov/ij/). Statistical analysiswas implemented using SPSS version 12 (Chicago, IL, USA).

Computational analysis in the microfluidic device

The chips were designed to induce a range of shear stressesin two-dimensional (2D) cell culture microchannels preparedwith microwell arrays. The chip design is similar to thosedescribed in previous studies30 (ESI† Fig. S2a). The dimen-sions of the cell culture chip were 1 cm × 3.5 cm, and theheight and width of the microchannel were 250 μm and500 μm, respectively (ESI† Fig. S2a). The shapes formed bythe complex flows (velocity and shear stress in the micro-fluidic channel) were modelled using the finite elementmethod (FEM) implemented in the CFD code (COMSOLMultiphysics 3.4, COMSOL Inc., MA, USA). A triangular gridsystem was employed for the calculation, and the number ofgrids was 18 752 (ESI† Fig. S2b). The viscosity of the mediumwas assumed to be the same as that of water (0.001 Pa s).

Lab Chip, 2014, 14, 4213–4219 | 4215

Lab on a ChipPaper

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

View Article Online

The shear stress in the microfluidic channel was calculatedbased on the results of the CFD analysis. The shear stress(τc), the stress imposed on cells attached to the microfluidicchannel surface, could be calculated according to

co

Vy

(5)

where V is the velocity of the fluid in the channel. The data wererecalculated based on the 2D simulation, taking into accountthe thickness of themicrofluidic channel (ESI† Fig. S2d). For thecomputational analysis, we performed the mesh sensitivity testof the microfluidic device (ESI† Fig. S2c). We found that the dif-ference of the velocity value is less than 1% between the meshsystems C and D (ESI† Fig. S2c). Finally, we used the mesh sys-tem ‘C’ for the CFD analysis (the number of elements is 18752).

Results and discussionCharacteristics of the yarn flow resistor (YFR)

In our preliminary experiments, the culture medium flow ratewas found to be approximately 60 mL h−1 through the micro-channel (250 μm high, 500 μm wide, and 30 mm long)

4216 | Lab Chip, 2014, 14, 4213–4219

Fig. 2 Quantitative analysis of the YFR flow characteristics. (a) The mean80 mm YFR (n = 8). (b) Plot of the mean flow rate as a function of the YFRrate in the microfluidic devices prepared using an 80 mm YFR with 250 aobtained from several types of micropumps. It shows the flow rate range ostatistical analysis was performed using the t-test. The error bars indicate th

without the YFR. The siphon pump flow rate needed to bereduced to physiological levels, which could be accomplishedusing the YFR. Fig. 2 shows the flow characteristics in themicrofluidic devices with the YFR. The siphon-induced flowrate in themicrochannel was dependent on the height of the res-ervoir, the length of the YFR, and the dimensions of themicrofluidic device. The flow rate increased proportionallywith the reservoir height. Fig. 2a shows the flow rate in amicrofluidic channel with a height of 500 μm preparedwith an 80 ± 3 mm YFR (n = 8). The flow rate clearlyincreased as the height of the fluid reservoir increased. Theflow rate was linearly proportional to the length of the YFR,as shown in Fig. 2b. This result indicates that the YFR lengthcould be a critical factor in the precise control of flow speedin microchannels. Fig. 2c shows the flow rate as a functionof the microfluidic channel height. An 80 mm long YFR (n = 8)was used, and the fluid reservoir height was set to 30 mm.Diverse pumping mechanisms can provide diverse ranges offlow rate, and Fig. 2d shows the different flow rates of thedifferent pumping systems. Pneumatic,23 piezoelectric,24

osmotic,3 scanning laser pulse,25 and electro-active polymer-driven1micropump systems have previously been used. The pro-posed YFR-based micropump produced a continuous flow rate

This journal is © The Royal Society of Chemistry 2014

flow rate as a function of the fluid reservoir height prepared using anlength for the initial 30 mm reservoir height (n = 8). (c) The mean flownd 500 μm channel heights (n = 8). (d) Comparison of the flow ratesf several micropumping systems including a YFR-based passive pump. Ae mean ± standard error (*P < 0.05).

Lab on a Chip Paper

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

View Article Online

(~102 μL h−1) and shear stress (~10−3–10−1 dyne cm−2), which is anattainable range for a system with a cheap and simple method(Fig. 2a, b, and d).

To date, several devices have been used to provide mediato microfluidic cell culturing devices, and conventionalsyringe pumps and roller-type pumps were typically used.Despite the diverse advantages of these pumps, their exten-sive applications are limited by the pump size and the powersources for integration to cell culturing incubator systems.Alternatively, several simple and cost-effective passive micro-pump systems have been developed to facilitate the cell cul-turing in an incubator.3,4,6,14 However, these have limits incontrolling the flow speed and in long-term operation. Theproposed YFR system overcomes these problems. In particu-lar, the fabrication process is simple and cost-effective, andthe flow rate and shear stress could be precisely controlledover a broad range.

Computational analysis of the shear stress and flow direction

Fig. 3 presents the computational results of the fluid flow dis-tribution in the microwell for cell culture under a controlla-ble shear stress. The simulation results illustrate that therecirculation flows in the microwell array were separatedfrom the main flow in the channel (white arrowheads), a typi-cal flow pattern observed in cavities26,27 (Fig. 3a). The

This journal is © The Royal Society of Chemistry 2014

Fig. 3 CFD analysis results obtained from the microfluidic system preparemicrowell array microfluidic channel. Red regions indicate higher flow ratethe white arrowheads. Lower flow rates were observed inside the microweof the velocity in the microchannel along the white dotted line shown in (a(d) Plot of the shear stress in the microfluidic channel. The data weremicrofluidic channel thickness (250 μm).

contour plot shows the magnitude of the fluid velocity in themicrofluidic device. The red area indicates a region of highervelocity relative to the blue area (Fig. 3a). The arrows indicatethe flow direction in the microfluidic device. A clockwise flowwas clearly observed in the microwells. The velocity distribu-tion in the microfluidic device is plotted as a function of they position in Fig. 3c (white dotted line in Fig. 3a). The insetshows the recirculation velocity in the microwell area inwhich the fluid velocity was 100-fold lower than the fluidvelocity in the main channel. Although the velocity of intersti-tial flow has not been elucidated, the order of interstitial flowhas been reported within 0.1–2 μm s−1.28–31 Thus, the flow ofthe YFR-based pump can mimic this interstitial flow leveland can be used for the study of cell responses under inter-stitial flow conditions. Fig. 3d plots the flow-induced shearstress in the microfluidic device. The data were qualitativelyand quantitatively similar to the velocity plots. The simula-tion results supported the utility of the microfluidic device asa cell-culturing platform that can generate a range of shearstress and flow velocities.

Response of the hMVECs to the flow controlled by the yarnflow resistance

As a proof of concept, we observed the response of the endo-thelial cells to the in vivo level shear stress. The flow pattern

Lab Chip, 2014, 14, 4213–4219 | 4217

d with an array of microwells. (a) Contour and streamline plots of thes (in the main channel). Separated recirculation flows are indicated bylls. (b) Arrows indicate the flow direction inside the channel. (c) Graph). The inset plots the velocity in the microwell area (orange dotted box).recalculated based on the 2D simulation with consideration of the

Lab on a ChipPaper

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

View Article Online

in the microfluidic channel imposed different levels of shearstress on the cells in the main channel and in the microwellareas over 1 week. The endothelial cell responses to differentshear stresses could be observed within a single microfluidicchip using the YFR flow control system. Fig. 4 shows the cellsattached in the microfluidic chip (height: 250 μm) under YFRflow control (3 mL per day). In the main channel (with ashear stress of 10−1 dyne cm−2), the attachment area andelongation of the cells were larger than the correspondingvalues measured in the microwells (with a shear stress of10−3 dyne cm−2) as shown in the ESI† Fig. S3. The hMVECalignment on the chip followed the flow streamlines in themain channel (Fig. 4a). The recirculation flow in the micro-wells appeared to affect the circular arrangement of hMVECs,although its flow rate is much smaller than that in the mainchannel. These results indicate that the endothelial cellsrespond to a wide range of slow fluid flow. Fig. 4b and eshow a quantitative analysis of the cell morphology underflow-induced shear stress or static conditions. In the mainchannel, the attachment areas and the cell perimeters werelarger than the values measured in the lower shear stress andstatic areas. The aspect ratios of the cells were also larger inthe main channel. The areas and perimeters of the cellsunder a low shear stress were similar to those measured

4218 | Lab Chip, 2014, 14, 4213–4219

Fig. 4 Flow-induced endothelial cell response in the microfluidic channemicrofluidic channel. The cells were aligned with the flow pattern in the chconditions. (c) The cytoskeleton arrangement under shear conditions or (dflow or static conditions. The statistical analysis was performed using the(*P < 0.05).

under static conditions (Fig. 4b). This result suggests thatthe interstitial flow only affects the direction of hMVECalignment. The cells grown under a low shear stress(10−3 dyne cm−2), however, were found to be more elongatedthan those cultured under static conditions (Fig. 4e). It is wellknown that the interstitial flow affects cell migration andinduces cell elongation.6,32,33 Fig. 4c and d show that thephysiological shear stress levels induced the arrangement ofactin stress fibers and altered cell shape. The cell shape andactin fiber patterns were similar to those reported in previoussheared EC studies.12,13 EC actin filaments grown understatic culture conditions were stained, revealing an arrange-ment along the flow direction. These results support thefact that physiological levels of flow can be regulated by theYFR-based passive micropump system which is more suitedto integrated microfluidic devices for long-term cell culturinglike organs-on-chips. The areas of the target hMVECswere larger than those measured among hMVECs culturedunder static conditions (Fig. 4a, c, and d). The cytoskeletonalignment in the hMVECs grown under shear stress culturingconditions coincided with the direction of the fluid flow.The YFR was useful for the long-term operation and theregulation of microscale fluid flow without the need for anyperipheral equipment. The regulated flow rate can

This journal is © The Royal Society of Chemistry 2014

l. (a) Comparison of the endothelial alignment and streamlines in theannels. (b) Quantitative analysis of the cell pattern under flow and static) under static culture conditions. (e) Aspects of the cells cultured underpaired simple t-test. The error bars indicate the mean ± standard error

Lab on a Chip Paper

Publ

ishe

d on

05

Aug

ust 2

014.

Dow

nloa

ded

on 1

0/08

/201

5 18

:30:

26.

View Article Online

consistently generate shear stress in a range that mimicsinterstitial flow. Such a range of shear stress levels canprovide cultured cells with a microenvironment similar toin vivo conditions, and the effect of shear stress to endothe-lial cell properties such as cell alignment, elongation andcytoskeleton alignment can be observed. The proposedsystem can be extensively used for the quantitative and quali-tative investigation of the shear stress effect on diverse cells.

Conclusions

We have developed a continuous flow-generating pump sys-tem using the siphon effect and YFR. The YFR could be easilyintegrated into a microfluidic channel, and the entire systemis simple, cost-effective and small enough to be integratedinto a microscale cell culture incubator. Our YFR passivemicroflow system can mimic in vivo conditions in an in vitromicroscale cell culture incubator. This application is muchbetter than the existing passive microflow pump systemsbecause it allows for precise control of the medium flow ratefor continuous, long-term microscale tissue culture. It willenable microfluidic cell culture systems to be much morebiomimetic, allowing the development of superior tissue cul-ture and organ on a chip platforms.

Acknowledgements

This paper is supported by the Korea Research FoundationGrant funded by the Korean Government NRF-2012R1A6A3A03039473.

References

1 G. H. Kwon, G. S. Jeong, J. Y. Park, J. H. Moon and S. H. Lee,

Lab Chip, 2011, 11, 2910–2915.

2 J. Y. Park, C. M. Hwang and S. H. Lee, Lab Chip, 2007, 7,

1673–1680.

3 J. Y. Park, S. J. Yoo, C. M. Hwang and S. H. Lee, Lab Chip,

2009, 9, 2194–2202.

4 O. C. Amadi, M. L. Steinhauser, Y. Nishi, S. Chung,

R. D. Kamm, A. P. McMahon and R. T. Lee, Biomed.Microdevices, 2010, 12, 1027–1041.

5 D. Huh, B. D. Matthews, A. Mammoto, M. Montoya-Zavala,

H. Y. Hsin and D. E. Ingber, Science, 2010, 328, 1662–1668.

6 W. J. Polacheck, J. L. Charest and R. D. Kamm, Proc. Natl.

Acad. Sci. U. S. A., 2011, 108, 11115–11120.

7 Y. J. Choi, S. Chae, J. H. Kim, K. F. Barald, J. Y. Park and

S.-H. Lee, Sci. Rep., 2013, 3, 1921.

8 S.-A. Lee, E. Kang, J. Ju, D.-S. Kim and S. Lee, Lab Chip,

2013, 13, 3529–3537.

This journal is © The Royal Society of Chemistry 2014

9 K. Yamamoto, T. Takahashi, T. Asahara, N. Ohura,

T. Sokabe, A. Kamiya and J. Ando, J. Appl. Physiol., 2003, 95,2081–2088.

10 H. Wang, G. M. Riha, S. Yan, M. Li, H. Chai, H. Yang, Q. Yao

and C. Chen, Arterioscler., Thromb., Vasc. Biol., 2005, 25,1817–1823.

11 T. Ahsan and R. M. Nerem, Tissue Eng., Part A, 2010, 16,

3547–3553.

12 S. G. Eskin, C. L. Ives, L. V. McIntire and L. T. Navarro,

Microvasc. Res., 1984, 28, 87–94.

13 C. G. Galbraith, R. Skalak and S. Chien, Cell Motil.

Cytoskeleton, 1998, 40, 317–330.

14 M. Rossi, R. Lindken, B. P. Hierck and J. Westerweel, Lab

Chip, 2009, 9, 1403–1411.

15 H. Chen, J. Cornwell, H. Zhang, T. A. Lim, R. Resurreccion,

T. Port, G. Rosengarten and R. Nordon, Lab Chip, 2013, 13,2999–3007.

16 M. Uematsu, Y. Ohara, J. P. Navas, K. Nishida, T. J. Murphy,

R. W. Alexander, R. M. Nerem and D. G. Harrison, Am. J.Physiol., 1995, 269, C1371–C1378.

17 S. Dimmeler, J. Haendeler, V. Rippmann, M. Nehls and

A. M. Zeiher, FEBS Lett., 1996, 399, 71–74.

18 P. J. Hung, P. J. Lee, P. Sabounchi, N. Aghdam, R. Lin and

L. P. Lee, Lab Chip, 2004, 5, 44–48.

19 S. R. Quake, Anal. Chem., 2007, 79, 8557–8563.

20 E. Berthier and D. J. Beebe, Lab Chip, 2007, 7, 1475–1478. 21 J. Ju, J. Y. Park, K. C. Kim, H. Kim, E. Berthier, D. J. Beebe

and S.-H. Lee, J. Micromech. Microeng., 2008, 18, 087002.22 J. Steigert, T. Brenner, M. Grumann, L. Riegger, S. Lutz,

R. Zengerle and J. Ducree, Biomed. Microdevices, 2007, 9,675–679.

23 J. Y. Kim, J. Y. Baek, K. H. Lee, Y. D. Park, K. Sun, T. S. Lee

and S. H. Lee, Lab Chip, 2006, 6, 1091–1094.

24 S. Ford and S. Soper, Analyst, 1998, 123, 1435–1441.

25 Y. Chen, T.-H. Wu and P.-Y. Chiou, Lab Chip, 2012, 12,

1771–1774.26 S. Hou, Q. Zou, S. Chen, G. Doolen and A. C. Cogley,

J. Comput. Phys., 1995, 118, 329–347.27 O. Botella and R. Peyret, Comput. Fluids, 1998, 27, 421–433.

28 S. R. Chary and R. K. Jain, Proc. Natl. Acad. Sci. U. S. A.,

1989, 86, 5385–5389.29 R. K. Jain, Annu. Rev. Biomed. Eng., 1999, 1, 241–263.

30 H. Dafni, T. Israely, Z. M. Bhujwalla, L. E. Benjamin and

M. Neeman, Cancer Res., 2002, 62, 6731–6739.31 M. A. Swartz and M. E. Fleury, Annu. Rev. Biomed. Eng., 2007,

9, 229–256.32 A. R. Wechezak, R. F. Viggers and L. R. Sauvage, Lab. Invest.,

1985, 53, 639–647.33 U. Haessler, J. C. M. Teo, D. Foretay, P. Renaud and

M. A. Swartz, Integr. Biol., 2012, 4, 401–409.

Lab Chip, 2014, 14, 4213–4219 | 4219